Identification of an Na_v1.1 sodium channel (SCN1A) loss-of-function mutation associated with familial simple febrile seizures

Results

Patients and Pedigree. The family that we studied (Fig. 1A) contained 35 individuals spread over four generations. A total of 13 living members had generalized tonic or tonic–clonic seizures with onset associated with fever, and none had FS beyond 6 years of age. None of the spouses had a history of seizures or familial epilepsy. The age at onset of FS ranged from 5 months to 4 years, with a mean age of 18 months. Subject I-2, who had already died at the time of the study, was reported to have had FS during early childhood. Most seizures were brief (1–5 min), but they lasted up to 15 min in patient IV-3. All affected members presented FS that fulfilled the criteria for simple FS (International League Against Epilepsy Commission; ref. 14). None of the patients had a history of meningoencephalitis or neurosurgery. The neurologic and psychiatric status of all 13 affected subjects was normal. Three individuals (III-3, IV-3, and IV-5; aged 38, 17, and 22 years, respectively) at the age of 10, 13, and 11 years, respectively, developed afebrile simple partial seizures of temporal lobe origin with vegetative or experiential phenomena. Very rare partial complex seizures or nocturnal secondary generalized tonic–clonic seizures also occurred in all of them. Patient III-8 was affected by posttraumatic epilepsy; a severe traumatic head injury at the age of 13 years accounted for afebrile seizures which started 1 year later at the age of 14 years. The electroencephalogram analysis showed interictal anteromesiotemporal epileptiform abnormalities that were seen in all three patients with afebrile temporal lobe seizures, whereas it showed right frontal sharp-waves in patient III-8 with posttraumatic epilepsy.

Electroencephalograms were normal in all other affected individuals, and they never revealed generalized epileptiform discharges. One patient (IV-3) had one typical seizure recorded with onset in the right midinferomesial temporal region. Two patients (III-3 and IV-3) with afebrile seizures had MRI evidence of unilateral mesial temporal sclerosis (MTS), which consisted of an abnormal high signal intensity detected on fluid-attenuated inversion recovery (FLAIR) and T2-weighted images ipsilateral with the electroencephalogram abnormalities. MRI study of patient III-8 revealed a posttraumatic scar in the right frontal region without any MRI evidence of MTS. MRI study was normal in an additional six affected individual (II-3, III-10, IV-1, IV-4, and IV-6) who had FS alone. Antiepileptic medication always allowed good control of afebrile seizures.

Thus, the pedigree analysis supports a highly (>90%) penetrant AD trait of simple FS in this family. Among the 13 affected family members, three patients developed subsequent afebrile seizures, whose electroclinical features were consistent with mesial temporal lobe epilepsy (MTLE) This phenotype resembles the phenotype that is encountered in families with AD FS, in which some individuals may develop later MTLE associated with MTS (15), whereas it differs from that reported in generalized epilepsy with FS plus (GEFS⁺) families (1).

Linkage and Sequence Analysis. We undertook linkage analysis in this family by defining the disease phenotype as simple FS. A genome screen revealed linked markers to the FEB3 locus on chromosome 2q23–24 that segregated with individuals who were previously classified as affected. Sequence analysis of the entire coding region of SCN1A Na channel gene revealed a T434C substitution (GenBank accession no. AY043484) within the coding region of exon 3, which alters an highly conserved methionine (Met145Thr, Fig. 2) within the S1 segment of domain I (DIS1) of the Na_v1.1 α-subunit (Fig. 1B). This mutation cosegregates with the disease because it was present in 12 family members with the affected haplotype and was not detected in all normal individuals from this family, indicating a disease-specific mutation; also, it was absent in 50 ethnically matched control individuals. Human Na_v1.1 consists of 2,009 aa and its sequence is highly conserved, with a 98% amino acid identity with the rat clone. Fig. 2 shows the homology of the amino acid sequences in the partial DIS1 of the human neuronal tetrodotoxin-sensitive channels (Na_v1.1, Na_v1.2, Na_v1.3, Na_v1.6, and Na_v1.7), one tetrodotoxin-insensitive channel (Na_v1.8), and rat Na_v1.1.

Functional Study. To study the functional effects of this mutation, we recorded with the whole-cell configuration of the patch-clamp technique the Na currents in tsA-201 human cells that were cotransfected with WT or FS mutant M145T hNa_v1.1 channels and hβ₁ accessory subunit. The first notable difference that we systematically observed was a 60 ± 2% reduction in M145T peak current density, as shown in Fig. 3, in which WT (199.1 ± 32.6 pA/pF) and M145T (79.9 ± 10.8 pA/pF) peak current densities have been plotted. The difference was statistically significant (P = 1.1 × 10^-4), suggesting an evident loss-of-function of the mutant. The threshold of activation and the time course of the Na⁺ currents showed considerable changes in the mutant. As shown in Fig. 4 A and B (same scales), Na⁺ currents started to activate at approximately -45 mV with the WT channel and at approximately -30 mV with the M145T mutant. Although this result should suggest a shift of the voltage-dependent open probability toward positive membrane voltages, we wanted to test whether the kinetics of opening were shifted in the same direction. Indeed, superimposition of the currents elicited at -10 mV (Fig. 4C) shows that M145T currents (○) are slower than WT currents (□), but if the mutated channels are tested at 0 mV (line), they open with the same speed of the WT channels tested at -10. We studied the activation kinetics in the voltage range from -30 to +20 mV and found statistically significant differences as shown in Fig. 4D, in which the current rise time (10–90% of the peak) is plotted vs. membrane potential. We observed a clear increase in the mutant (○), with a positive shift of 14.8 ± 1.3 mV (n = 9) in the voltage-dependence of the kinetics of activation. To clarify whether this effect is due to pure changes in the kinetics of the current or instead to modifications of the voltage-dependence of activation or inactivation, we examined the normalized peak conductance–voltage relationships and the voltage-dependence of steady-state fast inactivation. As shown in Fig. 5, we found that the effect observed in Fig. 4D can be reasonably explained by a simple positive shift of the voltage-dependence of peak conductance. In fact, the experimental data for WT (□, n = 11) and M145T (○, n = 14) channels can be fitted to Boltzmann curves with half-activation potential (V _½ values), which differ by an average of 10 ± 1.7 mV (see Fig. 5 legend), showing a positive shift for the mutant channel that is consistent with the previously observed change in the rise time of Na⁺ current activation. In contrast, the voltage-dependence of steady-state fast inactivation (▪, WT, n = 8; •, M145T, n = 15) did not show any significant difference.

In this study, we observed both a strong decrease in the Na⁺ current density and a significant positive shift of the voltage-dependence of activation. To investigate how the two effects could alter the all-or-none regenerative upstroke of the AP, we simulated both effects of the M145T mutation, independently and together, on a standard Hodgkin–Huxley model. The simulation is shown in Fig. 6, in which the superimposed traces of subthreshold responses (symbols) and APs (lines) are shown in control conditions (□,WT; AP threshold, -54 mV), for the M145T modified channel (▿; AP threshold, -27 mV), and with the functional biophysical effects described in Fig. 3 (○, only 40% of Na⁺ current, no voltage-dependent shift; AP threshold, -48) and in Figs. 4 and 5 (▵, only +10 mV right shift; AP threshold, -37 mV). It appears that the M145T channel produces an almost abortive AP, whereas the separate biophysical changes have much less severe consequences. However, these results concern a hypothetical cell with just Nav1.1 channels. Neurons express different Na⁺ channel isoforms, and the final effect of the mutation on the total Na current can vary according to the relative expression of Nav1.1 in a single neuron.

Discussion

We have shown that a missense mutation of hNav1.1 producing loss-of-function of the channel can cause simple FS. The phenotype of the present family differs from the clinical patterns encountered in families with generalized epilepsy with FS plus (GEFS⁺) phenotypes, which typically include FS, FS plus, and different types of generalized epilepsies (1). Except for patient III-8, who had afebrile seizures related to a severe traumatic head injury, three individuals had later afebrile, unprovoked seizures, whose electroclinical features were consistent with MTLE. Of these patients, two had a MRI-evidenced MTS, but only one (IV-3) showed FS lasting up to 15 min, indicating a putative example of complex FS. The association between familial AD FS and MTLE has already been proposed in large pedigrees, suggesting a common genetic basis whether MTS is present or not (2, 15). In particular, it has been shown that a family history of simple FS also can be a risk factor for MTLE (16).

To our knowledge, there is no unifying mechanism that can explain how the spectrum of the observed functional effects of epileptogenic mutations relates to the epilepsy syndromes seen in patients. Various epileptogenic mutations of SCN1A have been reported (17) that cause both loss- and gain-of-function of hNav1.1 channel. The mutation M145T that we describe here is a pathogenic mutation found in DIS1 of hNav1.1 that causes a single amino acid change; two other mutations in DIS1 (causing severe myoclonic epilepsy of infancy) have been reported (ref. 17 and supplemental material therein), but they cause the truncation of the protein and, thus, the production of a nonfunctional channel. The fact that Met-145 is conserved among both ortholog and paralog Na channels [which, thus, have both relatively low amino acid sequence identity (84%) and large distance in the phylogenetic tree (18)] suggests that the mutation is strongly pathogenetic, as described in ref. 17.

We have identified in patients affected by simple FS a mutation of Na_v1.1 that produces a decrease in current density and a right shift in the probability of opening, thus causing a loss of function. It is not clear yet how these effects can give rise to a network hyperexcitability that is able selectively to cause FS. The development of animal models and multielectrode array experiments would probably be useful in investigating this point. Moreover, because FS afflict a relatively large percentage of infants and children, it is likely that numerous other factors and genes can contribute to the generation of the heterogeneous seizure phenotypes observed in the FS spectrum.